TechVirtuoso

Storage Tiering vs Caching

November 12th, 2010 at 9:06 AM  3 Comments

During the first day of Tech Field Day #4 in San Jose we heard from two storage vendors who had two different thoughts on how storage should be configured.  We started out the day with NetApp who went over everything from their new OnTap 8.0.1 software, to the joint venture with Cisco and VMware called FlexPod as well as a presentation from Greg Kleimanon, Director of Marketing, on how NetApp uses caching in their arrays.

NetApp believes their Intelligent caching using Flash Cache is the best way to handle some of the hurdles that customers are facing today.  Kleimanon stated that Tiering is expensive and complicated to manage.  He did comment that there are companies (Like F5, another vendor from a previous Tech Field Day) that have created algorithms to dynamically manage tiering but he echoed that these algorithms are not proven.

The last presentation of the day was from Avere Co-founder/CEO Ronald Bianchini, Jr.  He was very passionate about his product and it showed through the presentation.  Avere uses your current “slow” storage for archive and uses a hybrid of RAM and SAS/SSD drives for faster reads/writes.  This solves the problem of speed in space by using a non traditional process of putting their box with the fast stuff (RAM & SAS/SSD) in front of your traditional SATA slow array.  It uses a algorithm to decide which items should be written/read from the RAM, the SAS/SSD drives or the SATA drives depending on the most efficient way to handle the data.

So what are your thoughts?  Caching or Tiering?

After seeing both presentations I think Tiering as implemented by Avere is the right way to implement the solution.  The way Avere breaks down the data and gives it the most efficient access path to the storage just makes sense.  Look for big things coming from Avere, especially in their FXT Series.

Update:  Here are the videos of the two presentations.  Watch and share your thoughts!

NetApp Presentation 1, 2, 3

Avere Presentation

Disclaimer:  Tech Field Day is organized by the great folks at Gestalt IT and paid for by the presenters of the event.  Even though my travel, meals and hotel accommodations were paid for my opinions are my own and it will not affect my posts.

  • I think there’s room for both. The difference, to me, is that tiering changes the authoritative location of data while caching makes a second non-authoritative copy that has to be kept consistent. Caching makes sense if the overhead of maintaining consistency is less than the savings in latency or reduced load on the origin. Tiering makes sense to optimize for performance on the origin, for the request stream that remains after caches have done their thing. They’re complementary, not mutually exclusive.

  • Startupgrrl

    What’s an algorithium?